利用1954～2001年我国150站日平均气温资料，分析了历年冬季期间（11月1日到次年3月31日）气温方差的变化与北极涛动的关系。发现北极涛动处于较强的正位相时，我国日平均气温方差偏小，反之则偏强。主要是通过影响西伯利亚高压中心区的海平面气压的高频变率而影响我国日气温的波动。季节尺度内异常低温事件（日气温距平<−2σ，σ为各站各冬季日气温距平的标准差）频次与气温方差之间存在反常的关系：温度方差大（小）时，异常低温事件频次下降（上升）。这与气温的非正态分布有关，也明显受北极涛动位相及强弱的影响。近50年来随着北极涛动指数的增强，西伯利亚高压中心区海平面气压方差显著下降（−10.7%/10a），我国日气温的变率也在显著减少（−4.1%/10a），而异常低温事件频次则显著增加（0.16d/10a）。

In the present study, the authors investigated the relationship between the Arctic Oscillation (AO) and the high-frequency variability of daily sea level pressures in the Northern Hemisphere in winter (November through March), using NCEP/NCAR reanalysis datasets for the time period of 1948/49-2000/01. High-frequency signals are defined as those with timescales shorter than three weeks and measured in terms of variance, for each winter for each grid. The correlations between monthly mean AO index and high-frequency variance are conducted. A predominant feature is that several regional centers with high correlation show up in the middle to high latitudes. Significant areas include mid- to high-latitude Asia centered at Siberia, northern Europe and the middle-latitude North Atlantic east of northern Africa. Their strong correlations can also be confirmed by the singular value decomposition analysis of covariance between mean SLP and high-frequency variance. This indicates that the relationship of AO with daily Sea Level Pressure (SLP) is confined to some specific regions in association with the inherent atmospheric dynamics. In middle-latitude Asia, there is a significant (at the 95% level) trend of variance of -2.26% (10yr)−1. Another region that displays a strong trend is the northwestern Pacific with a significant rate of　change of 0.80% (10 yr)−1. If the winter of 1948/49, an apparent outlier, is excluded, a steady linear trend of +1.51% (10 yr)−1 shows up in northern Europe. The variance probability density functions (PDFs) are found to change in association with different AO phases. The changes corresponding to high and low AO phases, however, are asymmetric in these regions. Some regions such as northern Europe display much stronger changes in high AO years, whereas some other regions such as Siberia show a stronger connection to low AO conditions. These features are supported by ECMWF reanalysis data. However, the dynamical mechanisms involved in the AO-high frequency SLP variance connection have not been well understood, and this needs further study.